the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Mar 2025
Connecting Deep Aquifer Recharge in California's Central Valley to Sierra Nevada Snowmelt via Multi-Sensor Remote Sensing Data
Abstract. California's arid Central Valley (CV) relies on groundwater pumped from deep aquifers (i.e., >50 m) and surface water transported from the Sierra Nevada to produce a quarter of the United States’ food demand. Similar to other basin aquifers adjacent to high mountains, the natural recharge to CV’s deep aquifers is thought to be regulated by the adjacent high mountains of the Sierra Nevada, but the underlying mechanisms remain elusive. We investigate large sets of geodetic remote sensing, hydrologic, and climate data and employ a first-order model assessment at annual time scales to investigate possible recharge mechanisms. Peak annual groundwater storage in the CV lags several months behind groundwater levels, suggesting a longer transmission time for water flow than pressure propagation. We further find that peak groundwater levels lag the Sierra Nevada snowmelt by about one month, consistent with an ideal fluid pressure diffusion time in the Sierra’s fractured crystalline body. Our results suggest that high mountain snowpack changes likely impact freshwater availability in the basin aquifers. Our analysis and a first-order pressure propagation model link the current precipitation and meltwater in the high mountain Sierra to deep CV aquifers through mountain block recharge process, highlighting the importance of longer groundwater flow paths through bedrocks for recharging deep aquifers in CV and other basin aquifer systems adjacent to mountains globally. This underscores the need for new hydroclimate models to fully account for the role of high mountains in lowland water cycles by including mountain block recharge, and revision of current management and drought resiliency plans in California.
- Preprint
(1989 KB) - Metadata XML
-
Supplement
(2531 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-4023', Anonymous Referee #1, 02 Jun 2025
The authors used a range of remote sensing, reanalysis and geodetic datasets to highlight the role of mountain block recharge (MBR) processes in the Central Valley aquifer system. I served as one of the reviewers of this manuscript when it was submitted to another journal, and unfortunately, many of the major comments raised at that time remain unresolved in this submission. These issues primarily relate to:
- Estimation of a lag time of approximately one month between Sierra Nevada snowmelt and groundwater level increases in the Central Valley;
- Overlooking the role of vadose zone processes in controlling groundwater recharge to the mountain aquifer unit;
- A failure to account for the influence of groundwater pumping in the interpretation of the results.
I provide several major comments below:
- Line 90-95- There is a misunderstanding of the conclusions presented in Armengol et al. (2024). That study showed that 31%–53% of groundwater in the southern Central Valley has the chemical signature of deep mountain block recharge (MBR). However, this does not imply that 50% of the total recharge in in the Central Valley originates from MBR.
- Lines 190-200- The authors stated that the “observation wells” in the southern Central Valley are less impacted by pumping. However, given the substantial volume of groundwater extraction in this region, the highest pumping region in the valley, it is difficult to believe that these wells are unaffected by cones of depression created by nearby pumping wells. A review of USGS wells from the climate response network (with no pumping impacts) reveals only a limited number of wells in California, with none in the Central Valley that are entirely unaffected by pumping.
- Lines 205-210 - Please check the screen intervals of the wells, as many wells in this region have long screens. As shown in earlier investigations in the valley by Armengol et al. (2024) and Thaw et al. (2022), such wells can reflect a mixed signal indicative of both shallow and deep aquifer recharge.
- Lines 295-300 - The authors state that pressure head propagation and mass flow occur over different time scales. However, it is incorrect to assume that an increase in pressure within the aquifer is not followed by mass flux. According to Darcy’s law, a change in pressure creates a hydraulic gradient, which in turn drives fluid flow.
- Equation 1. The authors assume that the fractured rock aquifer in the mountain aquifer behaves as a confined aquifer. Therefore, they use very small specific storage values. As authors know hard rock aquifers have low compressibility and specific storage values. However, these aquifers have secondary porosity due to fractures and may act more like an unconfined aquifer with higher specific yield that could deliver large amount of water due to changes in hydraulic head.
- Lines 310-315 - This conceptual model ignores unsaturated zone processes in the mountain block itself as it assumes snowmelt recharge is instantaneous.
- Line 328 – Can you justify the depth range of 600-1300 m. Does this correspond to the thickness of Central Valley aquifer?
- Lines 335-340 – Authors only consider saturated flow for the diffusion equation. This means that authors assume that the mountain aquifer unit is fully saturated and any snowmelt right away recharges the mountain aquifer. This is a major assumption and the delayed lag time caused by transport through the vadoze zone is important here.
- Line 654- It is strange that there is a 3-month delay between rises in groundwater level and GWS increase based on GRACE. This is physically impossible. Could this be related to processing of GRACE data?
References:
Thaw, M., GebreEgziabher, M., Villafañe-Pagán, J. Y., & Jasechko, S. (2022). Modern groundwater reaches deeper depths in heavily pumped aquifer systems. Nature Communications, 13(1), 5263. https://doi.org/10.1038/s41467-022-32954-1
Citation: https://doi.org/10.5194/egusphere-2024-4023-RC1 -
AC1: 'Reply on RC1', Susanna Werth, 05 Jun 2025
We thank the reviewer for their thoughtful comments. Below we provide responses to each comment, with the reviewer comments included in italic font and our responses in bold font.
The authors used a range of remote sensing, reanalysis and geodetic datasets to highlight the role of mountain block recharge (MBR) processes in the Central Valley aquifer system. I served as one of the reviewers of this manuscript when it was submitted to another journal, and unfortunately, many of the major comments raised at that time remain unresolved in this submission. These issues primarily relate to:
- Estimation of a lag time of approximately one month between Sierra Nevada snowmelt and groundwater level increases in the Central Valley;
We appreciate the reviewer's attention to the temporal dynamics of groundwater recharge. Our estimate of a ~1-month lag between peak snowmelt in the Sierra Nevada and subsequent groundwater level rises in the Central Valley is grounded in both a data-driven analysis (see manuscript Section 4.2 and Figure 6) and a modeling effort (see Section 4.3 and Figure 7). Specifically, our data-driven analysis indicates that peak groundwater levels lag behind the maximum water availability in the Sierra Nevada (which is composed of precipitation and snowmelt) by approximately one month. This is consistent with an ideal fluid pressure diffusion time in the Sierra’s fractured crystalline body. This finding aligns with previous studies that have documented similar lag times in mountainous regions with fractured bedrock aquifers (Meles et al., 2024; Wilson & Guan, 2004).
We will include these references in the revised manuscript.
- Overlooking the role of vadose zone processes in controlling groundwater recharge to the mountain aquifer unit;
We appreciate the comment. We emphasize that we are aware of the importance of the vadose zone and acknowledge that it plays a pivotal role in regulating groundwater recharge to mountain aquifer units (Meles et al., 2024; Wilson & Guan, 2004). Water infiltrating from precipitation or snowmelt must traverse this zone, where its movement is influenced by factors such as soil texture, moisture content, and the presence of fractures. In mountainous terrains, the vadose zone often comprises fractured bedrock and thin soils, which facilitate preferential flow paths that expedite the percolation of water to deeper aquifers. However, the heterogeneity of this zone can also lead to variable recharge rates, with some areas experiencing rapid infiltration while others exhibit significant delays due to lower permeability or increased evapotranspiration losses. Understanding the complex interplay of these processes for assessing recharge dynamics in mountain aquifer systems requires a multitude of interdisciplinary research efforts, which is well beyond the scope of this study.
In the revised manuscript, we will provide additional clarifications on the role of the vadose zone and its potential impact on the interpretation of our results, as discussed in the text.
We further recognize in the manuscript that the exact estimation of the pressure propagation time underlies multiple factors of uncertainties, like observational errors, data availability, spatial heterogeneities, and validity of assumptions (see lines 648-666, and 564-581). In addition, the focus of our manuscript is the data-driven analysis of a wide range of hydrogeological and hydrogeodetic data, considering the large temporal and spatial variability of the broader region over a two-decade period. Hence, estimating an exact propagation time is not feasible. Instead, the 1D-pressure propagation model ought to demonstrate the general feasibility of the MBR via a hydraulic connection from the Sierra Nevada to the deep Central Valley aquifers, and it does not include detailed physical process simulations. We emphasize this throughout the manuscript and call for detailed physical modeling in future studies (e.g., see lines 316, 640-647, 316).
Wilson, J. L., & Guan, H. (2004). Mountain-block hydrology and mountain-front recharge. In Groundwater Recharge in a Desert Environment: The Southwestern United States (Vol. 9, pp. 113–137). American Geophysical Union. https://doi.org/10.1029/009WSA08
- A failure to account for the influence of groundwater pumping in the interpretation of the results.
We recognize that groundwater pumping has a significant influence on aquifer dynamics in the Central Valley. The pumping in the valley increases the pressure gradient between the mountains and the valley to a larger extent than natural processes would. The goal of our work is to collect evidence for the existence of Mountain Block Recharge (MBR), rather than distinguishing between natural and anthropogenic processes.
The influence of pumping on the interpretation is addressed in detail in the manuscript discussed in lines 564-578. Pumping in the valley typically begins around April to May with the growing season and the majority of pumping in the valley occurs during the dry season (~May-June), which is also supported by the timing of the annual signal in GRACE groundwater storage change. Considering this for our investigation, the effect of pumping is likely to shorten the data-driven estimate for the timing of maximum groundwater levels (and thereby the delay between mountain water availability and groundwater levels). This, however, does not affect our conclusion that a delay exists; it only impacts the length of the delay. A longer “true” delay, masked by pumping influence, is also in agreement with model results for a range of lower diffusivities.
We will revise and expand the respective paragraph to make it more straightforward and provide further detail.
I provide several major comments below:
1. Line 90-95- There is a misunderstanding of the conclusions presented in Armengol et al. (2024). That study showed that 31%–53% of groundwater in the southern Central Valley has the chemical signature of deep mountain block recharge (MBR). However, this does not imply that 50% of the total recharge in in the Central Valley originates from MBR.
We thank the reviewer for this careful comment. We intended to emphasize the importance of MBR, which was highlighted by Armengol et al. (2024), who write:
“Mixing ratios show that 31%–53% of the groundwater system is recharged by Mountain Block Recharge originating from the Sierra Nevada. The high percentage of Mountain Block Recharge contribution indicates greater connectivity between the Sierra Nevada and the valley aquifer than previously thought (Meixner et al., 2016).”
We will adjust the statement to accurately reflect their results.
2. Lines 190-200- The authors stated that the “observation wells” in the southern Central Valley are less impacted by pumping. However, given the substantial volume of groundwater extraction in this region, the highest pumping region in the valley, it is difficult to believe that these wells are unaffected by cones of depression created by nearby pumping wells. A review of USGS wells from the climate response network (with no pumping impacts) reveals only a limited number of wells in California, with none in the Central Valley that are entirely unaffected by pumping.
We understand the concern regarding potential pumping influences. To mitigate the effect of pumping as much as possible, we selected observation wells located in areas with minimal pumping activity, based on data from the USGS Climate Response Network (see data lines 192ff and discussion lines 656-659 ). Selection criteria for observation wells include factors such as distance from known pumping centers and historical stability in water levels. While it's challenging to find wells entirely unaffected by pumping, our methodology aimed to minimize this impact.
We cross-referenced our findings with observations from across the valley through statistical analysis, highlighting spatial heterogeneity in the dynamics. We also include satellite-based measurements from GRACE, which provide integrated assessments of groundwater storage changes, offering another distinct insight into the system. The consistency of our findings across multiple data sources suggests that the observed signals are robust and not solely artifacts of pumping.
In addition, please see also our main remarks on this topic in response to your second general comment above.
3. Lines 205-210 - Please check the screen intervals of the wells, as many wells in this region have long screens. As shown in earlier investigations in the valley by Armengol et al. (2024) and Thaw et al. (2022), such wells can reflect a mixed signal indicative of both shallow and deep aquifer recharge.
Thank you for highlighting this important detail. We concur that well-screen intervals can influence the interpretation of groundwater data. In our analysis, we accounted for screen lengths by categorizing wells based on their construction details. We included the top perforation depth from CDWR and aquifer codes for USGS for classification of wells from deeper aquifer zones. This stratification enabled us to distinguish between signals from shallow and deep aquifer zones, thereby enhancing the accuracy of our interpretations.
In the updated manuscript (Section 2.2 on groundwater level data), we will provide a more detailed description of the well selection criteria and include information on perforation of well casings.
4. Lines 295-300 - The authors state that pressure head propagation and mass flow occur over different time scales. However, it is incorrect to assume that an increase in pressure within the aquifer is not followed by mass flux. According to Darcy’s law, a change in pressure creates a hydraulic gradient, which in turn drives fluid flow.
This is an incorrect characterization of the statement we provided. We clearly state that pressure gradient causes mass flux. However, a mass increase may lag due to a variable flux rate. Specifically, see line 297 in the manuscript:
“A pressure increase precedes the groundwater flow, which can initiate fluid diffusion from one location to another.”
5. Equation 1. The authors assume that the fractured rock aquifer in the mountain aquifer behaves as a confined aquifer. Therefore, they use very small specific storage values. As authors know hard rock aquifers have low compressibility and specific storage values. However, these aquifers have secondary porosity due to fractures and may act more like an unconfined aquifer with higher specific yield that could deliver large amount of water due to changes in hydraulic head.
Firstly, Equation 1 is a simple diffusion equation and does not explicitly assume confined or unconfined aquifer conditions. We agree, however, while these aquifers exhibit characteristics of both confined and unconfined systems, our model simplifies the system to facilitate analytical solutions. In the updated manuscript, we will better acknowledge this limitation and will expand our discussion of the implications of this assumption in the manuscript, suggesting that future studies could explore more complex representations (as already suggested in 640ff, for example).
Secondly, we do not directly apply specific storage values in our analysis or model. We discuss their impact on the physics of fluid flow, where the cross-sectional area and storage parameter have an effect, while pressure propagation remains independent of these parameters (L299-310). This derivation is generally valid, not only for the mountain block. In the revised manuscript, we will clarify this point.
6. Lines 310-315 - This conceptual model ignores unsaturated zone processes in the mountain block itself as it assumes snowmelt recharge is instantaneous.
This comment mischaracterizes our model. We do not assume an instantaneous model, nor does the model require such an assumption. The conceptual model merely summarizes the physical process linking deep valley aquifers to mountain aquifers. As explained above, the vadose zone acts as a buffer between surface water and the mountain block aquifer, and depending on its mechanical properties, can slow down or accelerate the recharge. Yet the physical process remains identical independent form the speed of recharge.
As stated above, in the updated manuscript, we will provide additional clarifications on the role of the vadose zone and how it may impact the interpretation of our results in the discussion section.
7. Line 328 – Can you justify the depth range of 600-1300 m. Does this correspond to the thickness of Central Valley aquifer?
No, this corresponds to an approximate average range for the length of the vertical hydraulic connection from the top of the Sierra Nevada Mountain block to the bottom of the Central Valley aquifer. We derived this from the dimensions and heights a.s.l. of the valley and the high mountains (disregarding extreme values), which are provided in Figure 2a, and considering that the western side is generally lower (900-2100 m a.s.l.). We also just noticed a spelling mistake in line 328, because we actually consider the depth range of 600-2000m (as shown in Figure 7). The approximate range derives as follows:
- Minimum value of 600 = 900 (mountains) – 400 (valley surface) + 100 (min depth of conf. units)
- Maximum value of 2000 = 2000 (mountains) – 400 (valley surface) + 400 (max depth of conf. units based on well depth)
In the revised manuscript, we will correct the error and make this description more straightforward in the respective text passage (current line 328) and add a reference to the figure.
8. Lines 335-340 – Authors only consider saturated flow for the diffusion equation. This means that authors assume that the mountain aquifer unit is fully saturated and any snowmelt right away recharges the mountain aquifer. This is a major assumption and the delayed lag time caused by transport through the vadoze zone is important here.
See response to main comments above on the role of vadose zone and diffusion equation.
9. Line 654- It is strange that there is a 3-month delay between rises in groundwater level and GWS increase based on GRACE. This is physically impossible. Could this be related to processing of GRACE data?
We have conducted a careful analysis of GRACE data, which included assessment only at its native spatial resolution (3 degree mascon tile). At that scale, the dataset is highly reliable, as established in the literature as a reliable regional observation of GWS (see references in lines 148 and 154). However, since any measurement comes with some uncertainty, we have conducted statistical tests (lines 472-477) and uncertainty analysis (lines 478-485, and Figs. S12, S13, 6g) to assess the robustness of the timing and significance of the delay estimates. Our findings suggest that uncertainty in the GRACE GWS is unlikely to alter our conclusions. In addition, the delay has been previously identified and confirmed by studies such as Argus et al. (2022) and Liu et al. (2019, Fig. 12), which we will incorporate into the revised manuscript.
We provided explanations of how this is possible, as outlined in the discussion lines 610-631. Our hypotheses are formulated using the best available data and insights, reflecting our current understanding and experience. It is crucial to consider that the Central Valley and Mountain aquifer systems are interconnected entities, linked via recharge paths such as the MFR and MBR (as illustrated in Fig. 2b and its accompanying caption). The recharge paths are long enough, but not too long, to discriminate between annual pressure and fluid flow progression, which affects each type of observation differently. Hence, it is further necessary to consider that while all presented datasets are sensitive to groundwater dynamics, the measurements from groundwater wells, vertical land motion assessed via GNSS or InSAR, and storage changes derived from GRACE utilize distinct sensor systems based on different physical principles, as we state in lines 610-616: “… the well level change is driven by changes in groundwater storage and pore fluid pressure, while the gravity-derived measurements only detect the change in mass, hence, storage changes”.
and second in lines 621-622: “The later peak in GWS might be primarily driven by annual storage variations in top unconfined aquifer layers…”.We will revise the respective discussion paragraph and better refer to Figure 2b, to make this explanation more straightforward.
When it comes to an integrated understanding of all the observations, we recognize that our interpretations may not be the final word on the matter. In contrast, we hope that our work stimulates a more detailed discussion, leading to the better integration of large-scale satellite-based observations with traditional ones, and ultimately to sustainable approaches for effective groundwater management.
Argus, D. F., Martens, H. R., Borsa, A. A., Knappe, E., Wiese, D. N., Alam, S., Anderson, M., Khatiwada, A., Lau, N., Peidou, A., Swarr, M., White, A., Bos, M. S., Landerer, F. W., & Gardner, P. (2022). Subsurface water flux in California’s Central Valley and its source watershed from space geodesy. Geophysical Research Letters. https://doi.org/10.1029/2022GL099583
Liu, Z., Liu, P., Massoud, E. C., Farr, T. G., Lundgren, P., & Famiglietti, J. S. (2019). Monitoring Groundwater Change in California’s Central Valley Using Sentinel-1 and GRACE Observations. Geosciences, 9(10), 436. https://doi.org/10.3390/geosciences9100436
References:
Thaw, M., GebreEgziabher, M., Villafañe-Pagán, J. Y., & Jasechko, S. (2022). Modern groundwater reaches deeper depths in heavily pumped aquifer systems. Nature Communications, 13(1), 5263. https://doi.org/10.1038/s41467-022-32954-1
Citation: https://doi.org/10.5194/egusphere-2024-4023-AC1
-
RC2: 'Comment on egusphere-2024-4023', Anonymous Referee #2, 11 Jun 2025
This manuscript proposes an innovative pressure diffusion model to identify mountain recharge for the deep aquifer systems of the Central Valley region of California. The idea of identifying flow paths between Mountain Front Recharge (MFR) and Mountain Block Recharge (MBR) – and their pathways to the unconfined and confined aquifers are novel. This data-based approach thus merits a serious discussion and consideration to be an addition to the vast research on California groundwater dynamics and trends.
The results are impressive and promising. However, the writing of the paper needs to be reoriented to make better sense and also provide future direction of the findings or the application of the proposed model.
The opening paragraph is inappropriate. It very abruptly moves to California in the second sentence of the paper without due introduction of the geographical context or the burning groundwater issues of the world or the region. The introduction section needs to specify the objectives of the study, and the target research questions in a detailed manner. The intellectual contribution of the manuscript is thus not clear in the current form.
The results should be better represented in something other than spatial maps. Time series of the groundwater flux or other water cycle components should be presented to explain the trends and seasonal cycles.
Similarly, the discussions and the results are currently combined into one long verbose section – but the main findings are not clearly apparent at the end. I would suggest the conclusions should be separated and the discussion section be more condensed.
I believe these changes would make the paper more succinct and easier to read.
Citation: https://doi.org/10.5194/egusphere-2024-4023-RC2 -
AC2: 'Reply on RC2', Susanna Werth, 13 Jun 2025
We thank the reviewer for the supportive evaluation and constructive comments. Below we provide responses to each comment, with the reviewer comments included in italic font and our responses in bold font.
This manuscript proposes an innovative pressure diffusion model to identify mountain recharge for the deep aquifer systems of the Central Valley region of California. The idea of identifying flow paths between Mountain Front Recharge (MFR) and Mountain Block Recharge (MBR) – and their pathways to the unconfined and confined aquifers are novel. This data-based approach thus merits a serious discussion and consideration to be an addition to the vast research on California groundwater dynamics and trends.
The results are impressive and promising. However, the writing of the paper needs to be reoriented to make better sense and also provide future direction of the findings or the application of the proposed model.Thank you for highlighting the novelty of the work as well as your interest in the results. We already have a very long discussion and tried to point out that coupled 3D flow and poromechanical modeling is required to clarify the details of the MFR/MBR link, as well as ideally also better data on the density and depth of fractures. However, we now see that this could be more detailed and strongly emphasized. We will revise the manuscript to improve the discussion and provide more details on future directions.
The opening paragraph is inappropriate. It very abruptly moves to California in the second sentence of the paper without due introduction of the geographical context or the burning groundwater issues of the world or the region. The introduction section needs to specify the objectives of the study, and the target research questions in a detailed manner. The intellectual contribution of the manuscript is thus not clear in the current form.
Thank you for pointing this out. We will revise the opening paragraph and formulate the objectives of the study and the target research question more specifically.
The results should be better represented in something other than spatial maps. Time series of the groundwater flux or other water cycle components should be presented to explain the trends and seasonal cycles.
I would like to explain our motivation for this choice thus far: Due to the sheer amount of observation points, using only time series to discuss the seasonal signal in the data is limiting the insight into the data. Showing individual time series will be either selective, or an overlay of many time series does make interpretation much harder. Since our study, mostly focusses on the timing (or phase) of the seasonal, it makes sense to aggregate the time series into this parameter, and plot it on maps, allowing an objective comparison of the results for all sites (and for observations from each sensor, incl. GNSS, storage components, GW levels, InSAR).
In the current manuscript, we already provide time series plots for well levels (Fig. 1a), GNSS VLM (Fig. 1 b), all storage time series (Fig. 1c), two detailed examples of the extraction of the seasonal signal from time series on well data (Fig. S7), and the final isolated seasonal signal from storages (Fig. 3h, also soil storage from different models in Fig. S12 and respective groundwater storage S13). The only dataset for which we do not provide insight into time series, thus far, is InSAR.
We understand that a few more examples would be helpful. Therefore, we suggest that we move Fig. S7 to the main manuscript to better highlight the process of extraction of the timing parameter. We will add time series examples for the InSAR data to Fig. 5. And we will plot more examples for time series (if visually feasible, for all) in supplemental Figures, to make the data more transparent.
Similarly, the discussions and the results are currently combined into one long verbose section – but the main findings are not clearly apparent at the end. I would suggest the conclusions should be separated and the discussion section be more condensed.
I believe these changes would make the paper more succinct and easier to read.Thank you for pointing this out. We will revise the manuscript to better separate the results and discussion, and to highlight the main findings more effectively.
Citation: https://doi.org/10.5194/egusphere-2024-4023-AC2
-
AC2: 'Reply on RC2', Susanna Werth, 13 Jun 2025
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 280 | 77 | 17 | 374 | 30 | 6 | 21 |
- HTML: 280
- PDF: 77
- XML: 17
- Total: 374
- Supplement: 30
- BibTeX: 6
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1